Manually Tracing Valve Curves

[rodabod]

I've tried searching, but couldn't find much useful information. I'm sure a few of you will have done this before.

I'd like to trace some triode-strapped pentodes manually. This as you might have guessed is for the purpose of some mic circuits. I don't want to approximate with pentode curves, especially since I'm more interested in the lower, less linear regions.

Can I simply used a fixed-bias circuit and sweep B+ from min. to max. connected directly to the plate, and repeat this for a series of fixed bias voltages? I was considering just using some diodes/LEDs or batteries for the bias supply.

 

[burdij]

Which ones were you thinking of doing. I did a couple of different EF86s and posted the results here:

http://www.groupdiy.com/index.php?topic=33789.0

I used a bench power supply with a variable C- and three DVMs.

The response was underwhelming 

 

[Gus]

Rodabod yes that should work.  I built a test circuit and posted about rp some time ago and I got a not so nice response so I removed it.

Burdij. I must have missed that post.

I have a mostly built test circuit variable B+, Bias etc. FWIW HF has sales and you can pick up a cheap DMM for little money. I payed $1.99 for a few I will check them against a Fluke and and make a correction chart for each one as they are used

 

[burdij]

Being a regular user of Circuit Specialists, I got a free DVM with each order so I have about a dozen. Battery goes dead, I open a new one (just kidding . . .). Unfortunately, they don't appear to send those out anymore. They do have a nice smd probe meter now though.

I set each C- value and then swept the B+ upward. When you get up into the higher current ranges on the mini-tubes, you will reach a point where the current will show a noticable rise per second. You don't want to go to current levels higher than that point for that grid voltage. I entered all the values in excel and then found a plotting program to produce the curves.

 

[rodabod]

Which ones were you thinking of doing. I did a couple of different EF86s and posted the results here:

http://www.groupdiy.com/index.php?topic=33789.0

I'm doing an EF14 and VF14. Yes, I know discussion of these valves has been done to death, but I want details that have not been discussed elsewhere.

Three DMMs? One for B+, one for bias... What's the other for? Heater?

On the topic of testing valves, have any of you guys tried to measure Rp manually? My method today was to AC-couple a potentiometer to act as a shunt across the plate, and adjust until the output level dropped by 6dB and then measure the pot value (at the same time taking the original plate load into consideration). Strange thing was that my values varied from speculation I'd heard from elsewhere by around 50%.... I know Rp can be calculated using other data, but some of that gets skewed when we operate the valves in a non-typical fashion.

 

[burdij]

Plate current is the third DVM.

 

[rodabod]

Another thing I've wondered about is general valve emission since this often decreases with age. My curves are less useful if they are out of spec.

Is there is a simple way I can test this for reference compared to a datasheet? I've only ever seen emission rated in mA which from what I understand is taken from when the valve is configured as a diode. Transconductance measurements would be tricker (if they exist) as no doubt they'd be given in pentode configuration.

People seem to often quote something along the lines of "reads X mA emission" but there is no mention of what voltage they are operating at. Is this test worthwhile, and if so, at what voltage?

 

[Kamel]

if it can help, there is a plethoric article in the bible "RDH4", chapter 3 page 90, about testing characteristics

 

[mad.ax]

Hi Rodabod,

you may like to download the last document at the end of this page:
http://www.jogis-roehrenbude.de/Roehren-Geschichtliches/Roe-Pruefer/Metrix_U61/Metrix-U61.htm

document:
http://www.jogis-roehrenbude.de/Roehren-Geschichtliches/Roe-Pruefer/Metrix_U61/metrix.pdf

(hint: EF14 is on page 77 of the pdf)

That is the combination listing for the Metrix U61. It was compiled by Metrix for what is often considered as the most accurate tube analyser ever made. For each tube type, it gives you the expected anode current for a given  Vg1, Vg2, Vg3, and Va. It also give you the expected mA/V characteristic.

Of course, you only get these characteristics for pentode mode if it's a pentode. But if the one you test shows the same results than listed, you can assume that it would pass the factory final inspection.

To make measurements in the same configuration than with an U61, all you need is a regulated 0~300V power supply, a small regulated variable DC PSU for the grid voltage, a milliampmeter, and a precise AC source for the heater.
I don't know if DC heater would give differents results from AC, but in the U61 it's AC
The HV must be regulated, that's the selling point of the U61. This way, when you set, say 200 VDC at the anode, it remains 200 V, no matter how much current the tube draws depending on Vg1...

Hope this helps
Axel

 

[rodabod]

Guys, thanks for the advice.

Axel, that's awesome for you giving me those suggestions. I'll need to try and get a 200V anode supply, but I can work on that.

Roddy

 

[mad.ax]

Here's a very simple 0~300V 100mA schematic wich should do the job nicely:

http://partselectronic.blogspot.com/2009/04/0-300v-adjustable-power-supply.html

Axel

 

[tubologic]

Guys, thanks for the advice.

Axel, that's awesome for you giving me those suggestions. I'll need to try and get a 200V anode supply, but I can work on that.

Roddy

For small signal (not power) tubes you could use a HEATHKIT IT-1121 Semiconductor Curve Tracer which can provide (plate) voltage sweeps up to 200V  and gate (grid) drive voltage steps from .05 V/Step to 1V/Step. All you need is to provide an external heater supply, and of course a scope to display the results. The IT-1121 is common and can be found very cheap.
Curves can also be plotted manually on graph paper,wich is a slow tedious and error prone process. Try to get a copy of the (excellent) book Tube Testers and Classic Electronic Test Gear by Alan Douglas which has a chapter about tube characteristics curve tracers  and detailed informations about how to build a cheap Transfer Curve Plotter and also how to use a HICKOK tube tester for curve tracing. All the informations you're looking for can be found in this book,and even more. I got the best results in terms of resolution and accuracy by using a METRIX laboratory tube analyzer (with built in regulated variable power supplies) coupled to a X-Y analog pen plotter. Since this gives me results in a printed form I no longer use my old Tektronix 570 or other methods for curve tracing.

 

[mad.ax]

The HEATHKIT IT-1121 is indeed a cheap curve tracer that could be usefull to trace curves... but I wouldn't trust it as an accurate measurement method. And it would be certainly useless as a 0~200V regulated variable DC supply, because of its operation principle. It provide voltage sweeps up to 200V really easily: 200V transformer>bridge rectifier.
And that's all. This way we get an half wave 240V peak at 50 or 60Hz. Pretty clever way to produce a curve on a scope screen.

But if one want to make accurate measurements on a curve, then a curve on a scope screen is not the easiest way...
The old guys use to hand draw the curve after making a lots of measurement with a lab analyzer like the Metrix
An X-Y analog plotter can speed up the process, but it needs to be carefully calibrated if the goal is to measure and not only compare...
Ultimately, a computerized tube tracer offer the best in terms of comfort, ease of use, and accuracy...

There's a very promising french project, but it's still under development...
http://www.dissident-audio.com/Traceur/Page.html
http://www.dissident-audio.com/Traceur/EN1.html

Axel

 

[mad.ax]

I'm bumping this thread because I've come across a really interesting website:

http://tubetesting.yolasite.com/

Lots of informations and links about tube testing, procedures, equipment, and so on...

Axel